City Leadership in the AI Era

Listen to host Stephen Goldsmith speak with two leading voices at the intersection of local government, data, and AI: Rochelle Haynes, Managing Director of What Works Cities and Carrie Bishop, who leads data and AI initiatives for the Government Innovation team at Bloomberg Philanthropies. Haynes and Bishop share advice for mayors on leading AI‑driven culture change, choosing meaningful use cases, and making data central to how modern cities solve problems. They explore how cities can use data and generative AI to move beyond traditional public meetings toward intentional, co‑created community solutions, featuring real examples and leadership advice.

Listen here, or wherever you get your podcasts. The following is a transcript of their conversation.

Stephen Goldsmith:

Hi. This is Stephen Goldsmith, professor of urban policy at the Bloomberg Center for Cities at Harvard University, with another episode of our podcast, a particularly cool podcast for Data-Smart City Pod because we have two of the country's most influential people in the area of cities and generative AI. One is a guest from Bloomberg Philanthropies, Carrie Bishop, and the other is Rochelle Haynes from What Works. So welcome to both of you.

Carrie Bishop:

Thank you.

Rochelle Haynes:

Wonderful. Thank you, Stephen.

Stephen Goldsmith:

So you both [have] interesting backgrounds in addition to what you do now. So just before we get into it, Carrie, why don't we start with you because I know you from your San Francisco days before you became a big deal at Bloomberg. So just tell us a little bit about what you did in San Francisco and what you're doing now in Bloomberg, and then we'll ask Rochelle to do the same thing.

Carrie Bishop:

Absolutely. Thank you so much for having me on. So in San Francisco, I was the chief digital services officer. I led a team for five years in San Francisco, helping the city digitize its 900+ services and built the team from, I was the first one there, so really built that team up and it's still going pretty strong today, which is great to see.

And then this role came along and I think it's just the perfect opportunity to understand how cities are using data at this critical moment in the history of using data in cities. And so it's really great to get to see not just the US, but more broadly as well across the globe, how cities are using data and AI. And so I lead a portfolio of initiatives that are geared towards helping cities improve their data and AI capabilities, apply those capabilities to really tough problems that cities are facing every day. And I work on the Government Innovation team here at the Philanthropy.

Stephen Goldsmith:

Thank you, Carrie. Rochelle, everybody's competing to be well certified by you. Tell us a little bit what you do at What Works.

Rochelle Haynes:

Sure. So What Works Cities, I'm the managing director of the program. I've been here for about three plus years, and I come by way of this work first as a native New Yorker. So I always point that out. And I think for me, growing up here in New York, I was keenly aware of how your zip code determined your experience in this city, whether it was the school you went to, what your infrastructure, your transportation looked like. And that just sparked an interest in me of wanting to be in public service.

And so prior to joining What Works Cities, I actually worked for 10 years in New York City government doing both affordable housing work, social services, and my last stint I was chief of staff of the city's homeless services department. And so I bring that lens to the work at What Works Cities, which is we are the international standard of excellence on what it means to be good, well managed local government. We're helping cities think through things from data governance, data management, their valuation, their contracting. It's all the nuts and bolts that helps a city run.

And we are here to provide free capacity building assistance to help them do that, but we also celebrate them. As Stephen mentioned, we offer certification at the silver, gold, or platinum level for the cities that you think are just best in class at doing good governance. And it's an opportunity to celebrate and recognize cities that are doing good work.

Stephen Goldsmith:

Between the two of you, that's quite a combination. I want to talk with you a little bit now on community engagement, but I want to talk about community engagement with a data and AI overlay. Like what has been community engagement and what could it be?

And maybe Rochelle, I'll begin with you. Let me give you my highly warped view of community engagement as a mayor and deputy mayor. You get your team of professionals together, you decide what policy is best for the community, everybody's well-intentioned, and then you go to a community meeting where people yell at you for a couple hours and then you go back and do what you were going to do anyway. So give us a version of community engagement that's a little bit more community and a little more engagement.

Rochelle Haynes:

I love the phrasing of that question, Stephen. And so for us, community engagement is about actually centering the community in a thoughtful and intentional way. It is part of, actually part of our What Works Cities standard, is for cities to engage the community. And when I think of it, I think of intentional engagement. So it's not having your plan already, but it's thinking about what are the accessible ways that you can connect with your residents and collect their feedback on their perceptions and experiences in their city. If you're going to invest in a new public park, what does the city actually want? What assets already exist within that park and what are the things that connect with cities?

And so I have an example that came to mind down in Savannah, Georgia, a city that is certified with us. They were about to make a major investment in their public parks. They were going to invest in communities that hadn't been invested in for over 20 years. Rather than, again, creating just the plan and saying to the community, "Here it is," they actually surveyed residents and asked them, "What would you like to see in this park? What is the realization of this? What would make you feel safe to come to the park? What would make you come here and connect with your residents and your community members more?"

And so they did a survey. They also did an inventory of physical assets that were in the park. And so they're looking at things from the benches to the playgrounds, the picnic tables, matched those two things together, and then came up with a plan. And the plan they came up with felt more accessible and functional for residents. They actually saw an increase in the number of youth engaging in the park, up by 37%. They saw more youth engagement in their public pool, which was actually a facility that wasn't being used as much. They were also able to identify things like seniors in the community needed meals delivered, and they were able to do that.

And so it was that intentional, thoughtful engagement up front, and then the physical mapping of assets and using data that helped them come up with a plan that was actually comprehensive that led to deeper engagement. They also saved some dollars because by mapping their assets, they were able to see where they could do some bulk buying during their procurement process for what they needed for their park system.

Stephen Goldsmith:

Carrie, I know you've watched a little bit our efforts with Santi Garces in Boston to look at changing the way generative AI could modernize performance management. Let me give you a community question that relates to that. So I'm also so old that I was around in the earlier days of Mike Bloomberg's open data initiatives. And in the very earliest days, that meant you put up whatever PDFs you had online and you said, "Hurrah, we have open data." Because that was it in the beginning. Then we have much more massive open data now.

So if you take what Rochelle talked about in terms of community engagement and you thought about what would generative AI and other AI tools help communities do with all that information? Chart us a course that utilizes generative AI to dramatically improve the quality and level of engagement between communities and cities.

Carrie Bishop:

Well, I too am old enough to remember when throwing any old thing up online and calling it open data was a thing. And I won't say we're necessarily beyond those years, honestly, even now. But one thing I do want to say is I think Rochelle gave us a great example of a city that is thinking about this work and intentionally using engagement with the community to develop something new.

But I think as well in addition to that, cities are already sitting on a gold mine of data that they already have. Think about all those open text fields in every survey that has ever been, every call to 311, every interaction in a public meeting, there is sort of troves of public sentiment already existing. Now of course, some of those channels are maybe not the broadest that they could be. We know that only a certain number and type of person shows up at a public meeting quite often, depending on when they're held. So we have to treat all of that with caution.

But I will say there's a ton of data already there and AI allows us to take that data and understand it, process it in a way that we've never really had the tools to do before. Usually the survey results come in, the free text box is there, somebody very keen will read it, but the rest of it goes on a shelf. So first of all, what do we already know? Because I think if there's one thing I've learned from doing community engagement work myself over the years, it's that communities really don't like it when you ask them the same question 50 times over and refuse to listen to the answers. So step one is what do we already know about our residents and how they're feeling, and how can we harness that knowledge and that intelligence that we already have?

I think for sure surveying residents is something that cities have been trying to do for better and worse over the years, and that's great. I personally have come to this work through a sort of design lens. So I think a lot about co-design and co-creation with communities. How do we actually encourage communities to participate? And so I think as cities thinking about all the different ways you might engage a community, there's everything from, yes, doing a survey or doing very formal structured public meetings, which I think have their place. And then there's also that more generative, how do we co-design solutions together with communities? I think that's really important.

One of my dream, maybe we are living in the future here, scenarios is that firstly, community organizations also harness the power of AI. And I think that's something that I would love cities to be thinking about. How do we encourage that? How do we encourage our community-based organizations to up their literacy around AI and grab a hold of these tools? Because I think it can help just as much as it can help cities.

And then how do community organizations come to cities and say, "Hey, look, we've trained a small model on how our community talks about this particular issue or that particular topic, and we would love you to use that and incorporate that in your work as a city as you start to think through what are the approaches it might take to," whatever policy area it is. That to me would be a dream scenario and a totally different way of communities to work with cities. It's like we are giving you through AI the tool you need to understand how we the community view this.

Stephen Goldsmith:

Let's stick with your answer for a little bit. Let me raise questions about it to you and Rochelle. So Rochelle, I heard Carrie say two different things, not inconsistent, but just two different things. One is involving the community in a design exercise. I was thinking augmented reality while she was talking and other virtual tools you could do. And the other was helping the community understand the data related to their neighborhood so they could better engage the city. And we've got a little project now at Harvard we're just starting on the latter.

I'm trying to figure out how you would help a community learn how to use generative AI so they can look at causation. So do we have more potholes in the next neighborhood? Why we do we have more? Do we have more drainage problems or is it the frequency of the services?

So how could we take Carrie's vision of the future of community engagement and help communities engage with the cities? What would that look like, where both sides are using generative AI? When I used to do community meetings when I worked for Mayor Mike, I did like six a week. It was always striking to me that some problem that we kept looking at, the community had a way to solve because they lived with the problem, they had context. So take us one step further in Carrie's vision of what would we do to advance this collaborative use of generative AI?

Rochelle Haynes:

Great question. And so I'm going to start with part of, when I think about the education, I think we are in an amazing moment actually right now. We've been talking about the digital divide for some time now. The What Works Cities standard tries to get at access and it was broadband, but there's a bigger question around making sure our communities, making sure our civil servants, our leaders, are ready for this transformation. This transformation, which Carrie always says, it's already here, we need to catch up. But it's really, no, it's really an exciting moment to be in.

And so this is a moment when I think that for generative AI, what it can do is democratize these processes. And I think that's what's extremely exciting about this period. And to help community members, residents get on board, I think we need to go to where people are at. And so how do we start to create spaces and put in spaces education and awareness around tools, really just hold workshops, libraries, spaces where there's youth. It can be an example for a great employment opportunity for youth and workforce development.

So how do you map out and think about what are the places where people are going to seek knowledge and information, and how do you then embed educational resources and tools related to generative AI? How do you think about, is there a digital economy that can be created out of this and become part of your economic development and workforce development plan? And so it's really getting keen on what are the spaces. And most places have this. There's spaces where people are going to get information. Public libraries still matter. So is it the public library? Is it the community center? Is it the local civic organization? And have that be the space where you convene folks and start that education process. And I would say, and at the same time, it's a moment to also do some training for your civil servants. And we do this hand-in-hand. Do a little train the trainer program in many ways. You train your staff to then go out and train the public on how to use the tools.

And there are great examples across the globe on how this has been done. Some interesting work I think has happened throughout Latin America in particular on this. I'll say Belo Horizonte is one of the cities that are in our network, and they identify WiFi and there were just gaps in access to WiFi. And so then they ended up surveying the community, having conversations with the community, and putting hotspots in particular places, and then also making sure it was accessible. So it's in three languages, Spanish, Portuguese, and English. And so thinking in those type of terms, and then also thinking about, hey, there's some people who actually don't have these resources and tools in their home. So getting the WiFi access is not enough. You also need to make sure there's centers that people can go and access.

And so when I think about this education perspective, it's thinking about what are the places where you can do the education and training, and then what is the goal of that? And I think it can be really interesting to tie it to economic development.

Stephen Goldsmith:

Yeah. Carrie, just continuing that for a second. You attended the last meeting of our chief data officers group at GovEx, and this is a great network of chief data officers from the largest cities, a nice partnership, particularly now that Oliver Wise is leading at Hopkins.

But you talked at that meeting a lot about AI literacy. Rochelle gave a nice answer for the community. Let's say you were the mayor of San Francisco, particularly interested in data literacy. Name two or three steps. What would you do to broaden data literacy inside the four walls of the city as an enterprise?

Carrie Bishop:

I agree, that is a really great convening. Our work generally spans beyond the US, but this particular network is the CDOs from the US and they really are the group that is in real time figuring all of this stuff out, and doing a great job with it, I must say. And yes, one of the things that came up at that convening was about data literacy or AI literacy. And every city around the world is thinking right now about how do we upskill our stuff around AI.

And happily, there is a ton of great content out there already specifically geared towards public sector employees. So that's really exciting. And I know through the What Works Cities network, there's content there that's available that cities can grab hold of. So there's lots of different ways to engage in this work.

I'm hesitating because it's very nuanced, this conversation, and there are some extremes going on at the moment. And I believe it will settle. But on one hand, extreme caution and fear around the risks around AI. And I think it's appropriate to think about the risks. That is something that should be top of mind for leaders and how you put just enough governance in there that can mitigate some of these risks. And on the other hand, we're seeing some examples of extremely optimistic, shall we say, deployments of AI.

And so we have this cycle going on, like is it really risky and we shouldn't do it? And then other people saying, "Well, let's ignore the risks and do it anyway." And it's created a bit of swirl. What I love about our programs, the City Data Alliance, the work Rochelle's doing at What Works Cities, is it's really like how do you apply this thinking to real world problems? And I think if you stay rooted in problem as opposed to just AI for everyone kind of without any real purpose, then that is a better way to stay grounded to the use case that you're trying to achieve. And then you can put the work and the things in place around it to make sure that succeeds.

And for staff, when you think about AI literacy, we want everybody to learn these things, but why? Why should I show up to this training? Why should I engage in this work? How will AI help me? Connecting it to something that will actually help staff move their work forward, that is the goal. You can start with use cases and put a lot of governance around thinking about which use cases make most sense for your organization.

There are other examples of cities doing some kind of sandbox type experiments where they create a safe space for experiments to run. I know Santi in Boston is doing a lot of that stuff. Seattle is also doing some great experimentation with the more cutting edge end of this technology, just to find out what will work. That's a really important stage in helping people understand the power of this technology.

We see other cities like San Francisco opened up Copilot for every single employee in the city. And so that's a different approach, but still it's bounded. It's like we've picked a particular tool, we're going to give everybody access to it, and then we're going to monitor and evaluate how that's being implemented and make some determination how that's going for the organization. I personally love the examples where cities are finding a specific problem they want to solve, figuring out is this the right technology to solve it with, and then doing some work to actually implement that.

Stephen Goldsmith:

Carrie, I've got a story I have to tell you, my favorite rat, R-A-T, story. So X years ago, I don't know, the earlier days of Bloomberg Philanthropies, they made a major gift to Chicago to set up a data analytics center. And a very talented woman, Brenda Berman, was in charge. I went over to see her one day and said, "How are you doing? She said, "We have a slow take up." And I said, "Well, why is that?" And she said, "Well, because a lot of the agencies, they don't have time to work on data. It's hard to get there."

So I said, "Well, what did you do about it?" And she said, "Well, I asked every agency," this is a little bit like what I did in New York. She said, "I asked every agency for a problem that they thought data would help solve."

And the first problem they selected was rats because the rat department, whatever that was called, wanted to know where the rats would be before the rats got there, not after the rats got there. Where were the lights bad? Where's the trash bag? And so my use case data story, to your point, is my rat story. Pick the problem and use the data to show how you can solve it. Now, I know your answer was much more sophisticated, but I had to take a second for my rat story.

Carrie Bishop:

That is exactly what we need, is we need these problems that data can help solve or that technology can help solve, and we just systematically need to go about solving them. I'm hoping what was learned in the rat scenario was that, hey, actually, if we use data and apply it to this issue, like, hey, what would it be like if we did that with this other thing we're having and this other thing we're having? What would it be like if we solved every problem this way?

Stephen Goldsmith:

Problem solving as a contagion.

Carrie Bishop:

From rats to contagion, sure.

Stephen Goldsmith:

Okay. Rochelle, would you lift us up a little bit here? Looking forward, you have a broad reach. You and Carrie have a remarkable reach into cities around the world. What piece of advice do you have for mayors that want to be the best in using data and AI to improve their community responsiveness and the results of those cities?

Rochelle Haynes:

I don't want to echo what's already been said, but it really is identifying what's a pain point in your city? What's a pain point in your community? Where I've seen some really interesting AI use cases is places where universally, from the city leadership to community residents, are like, "This is an issue and we need to figure it out."

I think of two housing related ones, probably my background, the housing ones always stay front and center in my mind, but in Mendoza, Argentina, they're using AI to detect illegal dumping in public spaces. A lot of these illegal dumping is happening in the favelas and poor communities. And so they're using AI detection software to be able to identify these illegal dumping sites and clear it up. What a health benefit, but also thinking about the public space.

I just heard about Newport News, Virginia has an issue with blight. Community residents were talking about how just blight is rampant throughout their communities and they wanted the city to do something. And rather than just saying it's an issue, it's also using AI software to go out and detect and do kind of proactive code enforcement. And so when we think about what we want cities to be doing, it's what cities in our network are actually doing, is they're saying, "Here's an issue we all can get behind. And is there an AI software that feels safe, reliable, ethical, and responsible that we can use to try to tackle this issue?"

And I think the more that we can uplift these positive examples of use cases and cities leaning in, I think more and more this will stop feeling scary and feel like the very practical tool we need to be able to govern our cities well, but to also be a way to actually engage our residents in, as Carrie mentioned earlier, that co-creation of the communities that we desire. And so my advice is to identify your problem, of course bring the data to it, but then get creative at how you solve it. And I imagine that there's enough tools and resources out there to help you be able to do that.

Stephen Goldsmith:

Carrie, final piece of advice, please.

Carrie Bishop:

For mayors who are leading on this agenda, what we're seeing is a real recognition that this is a leadership issue. It's not just a data nerds in the corner issue. It's actually fundamental to how a city operates in the modern era. And so the mayors that are really leaning into this, I'm thinking San Jose, Denver, in Latin America, Lujan de Cuyo, Recife, these are all cities where the mayors are deeply engaged in this work. They massively care about the outcome. They know that data is critical to achieving the goal, and they are acting as true leaders in driving culture change, clearing the pathway for these teams to operate.

Mayors that are cutting through all of that and saying, "No, the outcome is what matters, and we're going to forge a path through this," and making it a priority, the mayors that are asking for data in every single meeting, those are the mayors that are actually making this change happen. And so my advice to mayors is to think about this. This is a culture change and it starts at the top, and we stand ready to support mayors that want to take on that challenge. That is really my advice to mayors, is think of this as a leadership challenge and a culture change challenge.

Stephen Goldsmith:

Well, What Works and Bloomberg Philanthropies through its many contributions are making a huge difference for cities. So thanks for that. We've had a terrific conversation with Rochelle Haynes as the managing director of What Works Cities and Carrie Bishop, who leads data and AI initiatives for the Philanthropy itself on its government innovation program. Thank you so much to both of you for your contributions in our podcast and generally to cities.

Carrie Bishop:

Thank you so much.

Rochelle Haynes:

Thank you for having us.